Liquefaction of fine-grained soils from cyclic dss testing

Transcription

1 Liquefaction of fine-grained soils from cyclic dss testing John P. Sully and Ender J. Parra Principals, MEG Consulting Limited, Richmond, BC, Canada Gordon Fung and Haroon Bux Geotechnical Engineers, MEG Consulting Limited, Richmond, BC, Canada Paul A. Sully Laboratory Manager, MEG Consulting Limited, Richmond, BC, Canada ABSTRACT Generally accepted methods used to evaluate the potential for liquefaction have been developed primarily for clean sands or silty sands with up to about 35% fines content. As the fines content increases in sand, the in situ penetration test data is corrected in an attempt to compensate for the effect of fines content on the penetration resistance of the sand. Correction methods have been proposed for both SPT and CPT data. Where the silt content of sand is higher than 35%, correction factors do not appear to adequately reflect the increased resistance to liquefaction. Under these conditions, the generally accepted procedure is to base the assessment on laboratory tests. The paper presents the results of a series of cyclic simple shear tests performed on silty sands and silts from profiles in the Lower Mainland of British Columbia. The test results confirm that the increased resistance to liquefaction may not be adequately represented by the correction methods proposed for in situ penetration data. RESUMEN El enfoque normal para evaluar la resistencia a la licuacion se basa en la ejecucion de ensayos de penetration en campo. Sin embargo, se ha demostrado que la resistencia a la penetracion es sensible al porcentaje de finos en los suelos granular. Incrementos en la cantidad de finos causan una reduccion en la resistencia a la penetracion y una reduccion en la resistencia a la licuacion particularmente si el porcentaje de finos es mayor a 35%. La ejecucion de ensayos dinamicos de laboratorio es el metodo preferido para evaluar la resistencia a la licuacion para suelos finos. El articulo presenta resultados de ensayos dinamicos (DSS) realizados con arenas limosas y limos de la zona de Vancouver, British Columbia y la comparacion de resultados con el grafico basado en la resistencia a la penetracion. 1 INTRODUCTION The evaluation of the susceptibility of granular soils to liquefaction is commonly assessed based on in situ penetration test data. The method proposed by Seed et al. (1985) compares the cyclic resistance ratio (CRR) of the soil with the cyclic stress ratio (CSR) induced by the earthquake ground motions. The cyclic resistance ratio is defined based on either the N value from the standard penetration test or the tip resistance (q c) from a CPT sounding. The original chart presented by Seed et al (1985) is based primarily on the evaluation of case histories of liquefaction for clean sands. The chart presents a dividing line between cases where liquefaction was recorded and results where no liquefaction was evidenced. The dividing line is taken to be applicable to clean sands with fines contents less than 5%. Subsequently, lines corresponding to 15% and 35% fines content were added to the figure (Figure 1). The measured penetration resistance in granular soil is affected by the fines content of the soil. As the fines content of sand increases, the penetration resistance is decreased. For a sand with constant relative density (or void ratio), the measured penetration resistance needs to be corrected to compensate for the reduction due to the effect of the fines content. Considering that the soil cyclic resistance or strength is defined based on the penetration resistance, then the correction to the penetration resistance also applies to the soil cyclic resistance (Figure 2). The correction is directly related to the fines content, but the plastic component of the fines may also be a contributing factor (Figure 3). Ishihara (1996) suggests that the correction for fines content and plasticity would be of the form: (N 1 ) FC+PI = N 1 + N FC + N PI [1] A similar correction also exists for the use of CPT data in fine-grained soils. Figure 2 indicates the correction required for a sand of equal relative density where the penetration resistance is affected by the fines content. A possible correction factor on the cyclic strength of fine-grained soils for the effect of plasticity, suggested by Ishihara (1996) is indicated on Figure 3. Various approaches have been proposed for adjusting the penetration resistance (N or q c) to account for the influence of the fines content of the soils: Idriss and Boulanger (24) Seed et al. (23) Robertson and Wride (1998) Stark and Olson (1995) NCEER (1996 and 1998) Seed and De Alba (1986)

2 earthquakes in China. The Chinese Criteria suggest that fine-grained soils may be susceptible to significant strength loss if they satisfy the following conditions: % of the particles are finer than.5 mm liquid limit, LL < 35% ratio of water content (w) to liquid limit, w/ll >.9 2 max, l/ ' v max, l/ ' v at PI = Figure 1. Evaluation of CRR from SPT data from empirical liquefaction data (Youd et al. 21) Plasticity Index, PI Figure 3. Modification of cyclic strength of fine-grained soils allowing for the effect of plasticity index (Ishihara, 1996). Figure 4. N 1 and q c1 values as a function of fines content (Seed and De Alba, 1986). Figure 2. Definition of N 1FC for sands of equal relative density but differing fines contents (Ishihara, 1996). Typical correction values for both N 1 and q c1 as a function of fines content are indicated on Figure 4. However, it is commonly thought that the correction for fines content underestimates the real impact of the fines on the penetration resistance (ATC/MCEER 26). As a result, it is generally recommended that laboratory tests be performed on soils where the fines content may be large. Based on data published in the literature and our own experience, we consider that this should be the case wherever the fines content of a granular soil is larger than about 1-15%. Furthermore, we believe that laboratory testing is the most reliable way of determining the liquefaction resistance of silt. Empirical approaches have also been developed to evaluate the liquefaction susceptibility of fine-grained soils. Similar to the SPT approach for sands, the Chinese Criteria was developed from observations after Various modifications to the Chinese Criteria were subsequently proposed by other researchers. The Chinese Criteria were modified by Bray et al. (24) based on observations following the Kocaeli earthquake in Turkey. The basic conditions proposed by Bray et al. (24) are summarized on Figure 5. Boulanger and Idriss (24) suggest that if the plasticity index (PI) of the soil is less than 7, then the soil can be analyzed as if it were sand, using the penetrationbased approach. If the PI>7, then the assessment should be based on shear strength considerations. Based on recent results presented by Boulanger and Idriss (26), Bray and Sancio (24) and Sanin and Wijewickreme (26), the methodology recommended in the Greater Vancouver Task force report (27) suggests the following approach be followed: for PI<7, assume the material behaves like a sand and apply the penetration-based method to determine the cyclic resistance, OR undertake a

3 Plasticity Index, PI specific laboratory test program on good quality samples; for 7<PI<12, strain accumulation and postseismic settlement may be the primary concern and the material is considered less likely to liquefy. The post-seismic residual strength can be approximated by 8% of the static Su; for PI>12, the material is considered to behave as a cohesive material and the cyclic effects on stiffness and strength may be limited. Postseismic strength reductions are generally limited and Su is used for design. However, the task force urges caution in the application of these guidelines to sensitive and Application of the Bray et al. (24) criteria for liquefaction assessment of the overconsolidated liquefaction soils. susceptibility of Fraser River Delta silt Not Susceptible Wc / LL ML-1 ML-2 ML-3 ML-4 Moderately Susceptible to Liquefaction or Cyclic Mobility Susceptible to Liquefaction or Cyclic Mobility Figure 5. Application of the Bray et al. (24) criteria for liquefaction assessment for silt. 2 TEST PROGRAM Limited studies have been performed to assess the cyclic shear behaviour of fine grained soils (silty sands and silts). However, we believe that these soils are more resistant to initial liquefaction than inferred from fieldbased approaches. Laboratory testing approaches can be used to assess the liquefaction resistance of fine grained soils based on element tests on undisturbed or reconstituted samples. It is considered feasible to obtain undisturbed samples of acceptable quality in most fine-grained deposits. Where the fines content is sufficient to allow the recovery of good quality samples, it is generally possible to handle and prepare these samples for testing in the laboratory. A series of cyclic DSS tests have been performed on samples of silty sands and silts as part of projects performed in the Lower Mainland of British Columbia and other locations in North and South America. Only a portion of the results are presented herein. The testing was performed in the MEG Consulting Technical Services (MTS) Geotechnical Laboratory in Richmond, BC. MTS has two cyclic DSS testing machines from GDS Instruments in the UK. Both machines are rated to a testing frequency of 5 Hz, but cyclic tests are generally performed at 1 Hz. One machine has a vertical load capacity of 1 kn while the other is a 5 kn machine. The soil samples are contained in a series of low-friction rings during consolidation and shear, which is performed under constant-volume conditions. Samples tested have generally been recovered by thin-walled Shelby tube sampling. The Shelby tubes are scanned using gamma or X-rays to determine the areas of the sample in the tube that have been most affected by the sampling process. Samples for testing are only cut from parts of the tube where disturbance effects are not visible on the digital versions of the scans. Once the sample intervals to be tested have been defined, the sample tube is cut using a rotating tube cutter. Minimal pressure is applied to the tube to avoid deformations to the cross-section that may affect the sample. The sharpened disc cutter is rotated over the tube circumference to slowly cut into the wall. The tube is cut into sections about 75-1 mm long which permits the preparation of two or three hockey-puck sized samples for DSS testing. Once the tube is cut, the burrs (if any) at the ends of the cut section are removed and the sample is extruded in the same direction as the sample enters the tube during sample recovery in the field. Sample preparation is performed using a thin cheese wire to trim the sample height. The shear rings and internal membrane of the DDS equipment allow the 73 mm diameter sample to be placed directly into the membrane and rings with no trimming of the diameter required. Final sample dimensions are 25 mm high with a diameter of 73 mm. The testing program performed is generally broad and examines factors such as stress history and fines content on the response of fine-grained soils to cyclic loading. Since stress history is considered to be an important factor in characterizing the cyclic response, one-dimensional consolidation tests were performed on selected samples to determine the stress history of the deposit to be tested. To cover the range of conditions, tests have been performed on ko-consolidated samples without a static bias. In addition, the response after different loading histories (after initially identical stress conditions) has also been studied. All tests have been performed under constant volume conditions whereby the vertical load on the sample is automatically adjusted (computer controlled) to maintain no overall change in volume of the sample during shear. Cyclic loading for the set of tests presented in this paper has been performed under stress controlled conditions. Shear wave velocity measurements were made on selected samples at different times during the test by means of bender elements seated in the top and bottom caps. Once the cyclic loading was completed, either post-cyclic static shear or volume change were measured on selected samples. However, the information presented in this paper covers only the cyclic resistance results. The results presented in this paper are primarily from tests performed on good quality samples recovered in the field by thin-walled Shelby sampling. The highest quality sections of the Shelby tube have been identified

4 using gamma scanning and only these samples have been tested. An additional two tests are on samples prepared by moist tamping with another two samples prepared by dry tamping. The results presented here cover two aspects of the test program completed: the effect of fines content on the response of silty sand to cyclic loading, and the effect of the overconsolidation ratio on the resistance to liquefaction. 3 TEST RESULTS A summary of the samples considered in this paper and specific testing characteristics is presented on Table Fines Content The effect of fines content for the samples tested is presented on Figure 6. The test results presented on Figure 6 are from samples consolidated to pressures above those present in the field condition to ensure that no effect of overconsolidation history was present. Results presented are for soils that vary from clean sand with less than 5% fines content to silt with over 5% fines. The relative density of the sand samples was reasonably constant at around 45-55% to ensure similar field penetration resistances. The PI of the silt samples varied from non-plastic to a plasticity index of less than 12%. The results on Figure 6 are for failure defined by initial liquefaction or 3.75% single amplitude strain. The results on Figure 6 clearly indicate the increased resistance to liquefaction that occurs in samples with similar field penetration resistances as the fines content of the sample increases. For the case where N=15, the cyclic resistance ratio, given by the mid-point for each of the ranges, varies from about.1 for clean sand to just under.2 for silt. The increase in the resistance to liquefaction is clearly a function of the fines content, but would also appear to be dependent on the level of the cyclic stress ratio being applied (Figure 6). 3.2 Overconsolidation Ratio Sanin and Wijewickreme (26) have demonstrated the effect of overconsolidation on the resistance to liquefaction in silt. Testing was performed to verify the OCR effect for the design of ground improvement projects in the Lower Mainland. The testing program considers the effect of the above parameters separately and in combination. The stress-strain loops (for up to 1 cycles) for a typical sequence of tests are presented on Figure 7 for the silt sample ML-1 tested under OCR values in the range from 1. to 2.4. Testing was performed in the stress-controlled mode. All samples were saturated prior to testing. Since the saturation cannot be measured in the test, saturation was assumed to be complete after passing two pore volumes of distilled/de-aired water through the sample under a nominal backpressure and conditions of constant volume. Figure 6. Results of cyclic stress-controlled DSS tests on sands with differing fines contents (F c). The cyclic stress ratio for all tests is ; the first and last loops are highlighted on the figure. The change in the vertical stress ratio (equivalent to the pore pressure induced during cyclic loading) is presented for the same three tests on Figure 8. All samples indicate a similar final pore pressure ratio of about 7% but at widely differing numbers of cycles. The number of cycles to failure increases dramatically as the OCR of the sample increases. For an OCR=1, the sample achieved 5% strain very quickly after only 4 cycles. The numbers of cycles increased to 26 and 1 as the OCR increased to 1.7 and 2.4, respectively. The shape of the stress-strain curves also changed noticeably (Figure 7). As indicated on Figure 8, after completing the first cycle, the rate of pore pressure generation with number of cycles is approximately equal for OCR=1 and 1.7, but decreases significantly for OCR=2.4. A comparison of the results obtained for samples with differing OCR values is presented on Figure 9. The lowest curve corresponds to normally consolidated clean sands with similar field penetration resistances. The remaining three ranges correspond to soils with fines contents greater than 5%, plasticity indices in the range non-plastic to 11% and overconsolidation ratio increasing from OCR=1 to OCR=2.4. for the clean sand, at N=15 cycles the average cyclic resistance ratio is about.11. This increases to.18 as the fines content increases to more than 5% (consistent with Figure 6). As the OCR increases to 1.7, for N=15 the CRR increases to.35, while for OCR=2.4 the range is asymptotic to a value of about N=3. Figure 9 confirms the significant effect that the OCR may have on the soil resistance to liquefaction. The results on Figure 9 are for failure defined by initial liquefaction or 3.75% single amplitude strain.

5 Vertical Stress Ratio SHEAR STRESS (kpa) SHEAR STRESS (kpa) SHEAR STRESS (kpa).26 stress ratio ( cyc / vc 1 Hz for 4 cycles CSR = OCR = 1. No. of Cycles = characteristics. The post-cyclic response would also appear to be more representative of a material with a higher PI, since the post-cyclic strength loss was determined to be negligible for the soils tested st Loop Main Loop Last Loop CSR = OCR = 1.7 No. of Cycles = stress ratio ( cyc/ 1 Hz for 26 cycles SHEAR STRAIN (%) st Loop Main Loop Last Loop stress ratio ( cyc/ 1 Hz for 1 cycles SHEAR STRAIN (%) CSR = OCR = 2.4 No. of Cycles = V/ ' VC SHEAR 4 STRAIN 5(%) consolidated 1 to different OCR values CSR = Number of Cycles 1st Loop Main Loop Last Loop Figure 7. Stress-strain loops for sample ML-1 OCR = 1 OCR = 1.7 OCR = Figure 8. Variation of vertical stress ratio with number of cycles for sample ML-1 with different OCR values It is interesting to compare the results from the cyclic DSS laboratory tests with the method proposed by Bray et al. (24). The index parameters for the silt samples ML-1, ML-2, ML-4 and ML-4 are plotted on the Bray et al. (24) chart indicated on Figure 4. The index properties are summarized in Table 1 below. The Bray et al. chart would suggest that these samples are potentially susceptible to liquefaction or cyclic mobility. The results of the cyclic DSS tests would suggest that cyclic mobility or strain accumulation is more likely for a soil with these Figure 9. Results of the DSS tests on the cyclic strength of fine-grained soils under different overconsolidation ratio (OCR) 4 CONCLUDING COMMENTS The results presented above clearly indicate the increased resistance to liquefaction caused by an increase in fines content of sand and the beneficial effect of OCR for soils with high fines content, even if nonplastic or of low PI. The trend in the increased resistance can be visualized by comparing the results of the cyclic DSS testing with the data provided by the Youd et al. (21) liquefaction resistance curves (Figure 1) for the differing fines contents. Figure 1 is drawn for a M=7.5 earthquake which corresponds to 15 equivalent cycles of shaking. In order to generate the appropriate ordinates for other numbers of cycles, the magnitude scaling factors proposed by Idriss (29) have been applied to Figure 1. The comparison is indicated on Figure 1. The results of the cyclic DSS tests are indicated as solid or dashed lines, depending on the fines contents. The shaded areas on Figure 1 correspond to the three fines contents considered on the Youd et al (21) chart (Figure 1). For the clean sand shaded area at the bottom of Figure 1, the lower and upper ranges correspond to a variation in (N 1) 6 of 5-1. The numbers of cycles, N, vary from 4 for a M=6 earthquake to 24 for an M=8.2. The Youd et al. (21) chart on Figure 1 is for M=7.5 with N=15. Over the range from N=4 to N=24, the CRR from the laboratory tests on clean sand (Dr = 45-55%) agree reasonably well with the field data. At low numbers of cycles, the DSS data are slightly lower than the data

6 provided by Youd et al., although the DSS data provide progressively higher resistances as the number of cycles of shaking increases. The differences in response are likely due to the contractive nature of the moist-tamped sample preparation method. Overall, the comparison between the field and laboratory results for clean sand is good. At the upper end of the scale, the cyclic DSS results for the normally consolidated silt coincide with the upper limit for the Youd et al. fines content adjusted field curve. However, notably the cyclic resistance increases rapidly as the OCR increases. The effect of OCR would appear to be significant than the effect of fines content. Overall, the laboratory and field liquefaction resistance curves are in reasonable agreement. This is not surprising since the field curves are based on measured/interpreted response during/after earthquake events. This does not change the earlier comment that the fines corrections applied for SPT and CPT data in finegrained soils do not adequately account for the effect of fines content. At numerous sites, we have consistently witnessed that the (N1)6cs values in sands with silt contents of 15-3% or more, indicate higher susceptibilities to liquefaction than indicated by cyclic DSS tests performed on recovered samples. We are presently evaluating the results to pursue a more appropriate fines content correction. This conclusion would suggest that the N 1 that is determined from the Seed et al. curve could be considered as a lower bound value for the correction of (N 1) 6 to (N 1) 6cs in order to compensate for the effect of the fines content. Robertson, P.K., and Wride, C.E Evaluating cyclic liquefaction potential using the cone penetration test. Can. Geo. J., Vol. 35, No.3, pp Sanin, M.V. and Wijewickreme, D. 26. Cyclic shear response of channel-fill Fraser River Delta silt. Soil Dynamics and Earthquake Engineering, vol 26, No. 9, pp Seed, H.B. and De Alba, P Use of SPT and CPT tests for evaluating the liquefaction resistance of sands. Proc. In-Situ Test, ASCE, pp Seed, R.B. et al. 23. Recent Advances in Soil Liquefaction Engineering: A Unified and Consistent Framework. Keynote Presentation, 26 th Ann. Conf., ASCE, Long Beach, Calif, USA. Stark, T.D. and Olson, S.M Liquefaction resistance using CPT and field case histories. J. of Geot. Engng., ASCE, 121(12), Youd, T.L., et al. (21). Liquefaction Resistance of Soils: Summary Report from the 1996 NCEER and 1998 NCEER/NSF Workshops on Evaluation of Liquefaction Resistance of Soils. ASCE, Journal of Geotechnical and Geo-Environmental Engineering, October, pp REFERENCES ATC/MCEER 26. Liquefaction Study Report and Design Examples, LRFD Guidelines for the Seismic Design of Highway Bridges, ATC/MCEER, Univ.of NY. Boulanger, R.W. and Idriss, I.M. 24. Evaluation of potential for liquefaction or cyclic failure of silts and clays. Report No. UCD/CGM-4/1, Univ. Califonia at Davis. Boulanger, R.W. and Idriss, I.M. 26. Liquefaction susceptibility criteria for silts and clays. Journal Geot. Geoenv. Engng., ASCE, 132(11): Bray, J.D. et al. 24. Liquefaction susceptibility of finegrained soils. Proc., 11 th Int. Conf. Soil Dyn. & Eq. Eng., Berkeley, Calif., Vol. 1, Greater Vancouver Regional Task Force Report. 27. Geotechnical Guidelines for Buildings on Liquefiable Sites in Accordance with NBC25, 74 p. Idriss, I.M. 29. An update to the Seed-Idriss simplified procedure for evaluating liquefaction potential. Proc. TRB Workshop on New Approaches to Liquefaction, Pub. No. FHWA-RD , January. Idriss, I.M. and Boulanger, R.W. 24. Semi-empirical procedures for evaluating liquefaction potential during earthquakes. Proc., 11 th Int. Conf. Soil Dyn. & Eq. Eng., Berkeley, Calif., Vol. 1, Ishihara, K Soil Behaviour in Earthquake Geotechnics, Oxford University Press Inc, New York, United States. Figure 1. Comparison between DSS results and cyclic resistance curves recommended by Youd et al. (21)

7 Table 1. Summary of properties for DSS tests Soil Type Description Depth ' vo In-situ OCR (1) PI ' v test OCR test (2) Fines (%) Relative Density, Dr (%) Sample preparation method SP - 1 Fine SAND with mica Dry deposition, tamping SP - 2 Fine silty SAND Dry deposition, tamping SP - 3 Fine SAND Moist tamping SP - 4 Fine SAND Moist tamping SM/ML - 1 Sandy SILT Non-Plastic undisturbed sampling SM/ML - 2 Interbedded sandy SILT to silty sand Non-Plastic 85 - undisturbed sampling ML - 1 SILT with some clay and trace sand undisturbed sampling ML - 2 Clayey SILT with some sand undisturbed sampling ML - 3 Clayey SILT with organics undisturbed sampling ML - 4 SILT with some sand undisturbed sampling Notes: (1) OCR interpreted from field and laboratory test results (2) OCR imposed on cyclic DSS samples